Int. J. Environ. Res. Public Health 2012, 9, 3465-3483; doi:10.3390/ijerph9103465 OPEN ACCESS
International Journal of Environmental Research and Public Health ISSN 1660-4601 www.mdpi.com/journal/ijerph Article
Socio-Environmental Factors Associated with Self-Rated Oral Health in South Africa: A Multilevel Effects Model Bukola G. Olutola 1 and Olalekan A. Ayo-Yusuf 2,* 1
2
School of Health Systems and Public Health, University of Pretoria, Pretoria 0002, South Africa; E-Mail:
[email protected] Department of Community Dentistry, School of Dentistry, University of Pretoria, Pretoria 0002, South Africa
* Author to whom correspondence should be addressed; E-Mail:
[email protected]. Received: 24 July 2012; in revised form: 17 September 2012 / Accepted: 25 September 2012 / Published: 2 October 2012
Abstract: Aim: This study examined the influence of the social context in which people live on self-ratings of their oral health. Method: This study involved a representative sample of 2,907 South African adults (≥16 years) who participated in the 2007 South African Social Attitude Survey (SASAS). We used the 2005 General Household Survey (n = 107,987 persons from 28,129 households) to obtain living environment characteristics of SASAS participants, including sources of water and energy, and household cell-phone ownership (a proxy measure for the social network available to them). Information obtained from SASAS included socio-demographic data, respondents’ level of trust in people, oral health behaviors and self-rated oral health. Results: Of the respondents, 76.3% self-rated their oral health as good. Social context influenced women’s self-rated oral health differently from that of men. Good self-rated oral health was significantly higher among non-smokers, employed respondents and women living in areas with higher household cell-phone ownership. Furthermore, trust and higher social position were associated with good self-rated oral health among men and women respectively. Overall, 55.1% and 18.3% of the variance in self-rated oral health were explained by factors operating at the individual and community levels respectively. Conclusion: The findings highlight the potential role of social capital in improving the population’s oral health.
Int. J. Environ. Res. Public Health 2012, 9
3466
Keywords: self-rated oral health; social capital; mixed-effects model; cell-phone ownership; trust; South Africa
1. Introduction Self-rated health refers to the perception an individual has about his/her health. This perception has been directly associated with the individual’s experiencing physical symptoms [1]. Self-rated oral health is commonly regarded as a global rating of oral health-related quality of life; as it takes into account both the physical and psychological dimensions of health [2]. In particular, people use discomfort in the mouth, the inability of the mouth to function properly, as well as the effect of adverse oral health on social interactions in assessing their own oral health. Factors such as higher education [3], current employment [3,4], being of non-African American race [5,6], higher income [4,7], female gender [4,5] and younger age [5] have all been positively associated with good self-rated oral health. Self-rating of oral health as being good has also been positively associated with a higher number of retained teeth and a recent visit to a dental professional [8]. In the 2003/2004 South African Demographic and Health Survey (SADHS), 16% of the respondents reported having had problems with their mouths and/or teeth in the six months preceding the survey [9]. Self-reporting of problems with the mouth and/or teeth varied between ethnicities/race groups and provinces [9]. This geographical variation in self-reported oral health outcomes could be related to variations in the cultural beliefs and traditions, education levels, access to and quality of dental services and other socioeconomic characteristics of the individuals and their environment. A study by Turrell et al. [10] revealed that there was a significant positive association between a neighborhood’s socioeconomic characteristics and self-rated oral health, irrespective of an individual’s socioeconomic status. Variations in self-rated oral health have been attributed to differences in many socioeconomic and demographic factors which may be observed at both the level of the individual and the neighborhood or community [10]. Most of the prior studies, conducted mostly in developed countries, have focused on individual-level socioeconomic status as determinants of oral health. Thus far, only limited information is available on the influence of neighborhood socioeconomic characteristics on the oral health of South Africans, even though addressing socioeconomic disparities in health resulting from decades of apartheid rule [11] remains a public health priority in South Africa. The aim of the present study was therefore to examine the influence of the social context in which people lived on their self-rating of their oral health, independent of personal risk indicators for good oral health. 2. Materials and Methods 2.1. Data Source The individual-level characteristics such as self-rated oral health, oral health behaviors and the socio-demographic characteristics of participants in this study were obtained from the 2007 South African Social Attitudes Survey (SASAS), while the area-level characteristics, outlined further on in
Int. J. Environ. Res. Public Health 2012, 9
3467
this paper were obtained from the General Household Survey (GHS) conducted in 2005. The master samples of both the SASAS and GHS datasets consist of enumeration areas, which are the smallest geographical units that make up municipalities in South Africa. The municipalities (the equivalent of US counties) are the lowest level of government administration and service delivery, hence, they are likely to have meaning for and be significant in terms of where the study participants reside with regard to potential interventions that can be focused on environmental factors that may influence their oral health. Therefore, the two datasets were linked at the municipality level through similar codes that were uniquely assigned to each municipality in the two datasets. A new dataset was then created to form the basis of analysis in the current study. The 2007 South African Social Attitudes Survey (SASAS) was a nationally representative sample of adults 16 years and older. Participants were selected following a multi-stage probability sampling strategy. A sample was drawn from the master sample of the South African Human Sciences Research Council (HSRC), which consisted of 1,050 enumeration areas drawn from the 2001 Census. From each of the enumeration areas, 10 visiting points were randomly selected, resulting in a total of 10,500 visiting points in the master sample. The enumeration areas were stratified to be representative of the socio-demographic domains of the province(s), geographical sub-types (tribal areas, formal rural, formal and informal urban) and the four population groups—black, white, coloured (of mixed race) and Indian/Asian—in South Africa [12]. However, for the 2007 SASAS, a total of 4,000 households/visiting points were randomly selected from the master sample. Each person was then randomly selected from each household, without replacement. Efforts were made to secure an interview with each selected person by making three visits before registering the person as non-responding. A sampling weight which took into account the response patterns was applied to produce a representative sample of South Africans ≥ 16 years. In the 2005 GHS, multi-stage stratified samples were drawn from Statistics South Africa’s master samples, which, like the SASAS samples, were taken from the enumeration areas established during the 2001 Census 2001 for the 2005 GHS. The detailed methods used in ensuring standardized data collection, interviews and consent procedures for the GHS have been previously published [13]. The response rate for the 2007 SASAS was 72.6% (n = 2,907). For the 2005 GHS, the response rate was 87.5% of the targeted 32,146 households (n = 107,987 individuals). The very large sample for the GHS thus provided a unique opportunity to compute area-level characteristics for the corresponding municipalities where the participants of the 2007 SASAS lived (average n = 121 households per municipality). However, two of the municipalities from the 2007 SASAS could not be merged with the 2005 GHS data, because there were no corresponding municipalities in the 2005 GHS. This reduced the sample size by 4%, giving rise to n = 2,791. 2.2. Dependent Variable Self-rated oral health was assessed by the question “How would you rate your oral health?” and respondents were asked to pick one of the following options: “very good”, “good”, “neither very good nor good”, “poor” and “very poor”. Following the approach used in similar studies [5,10], the options were dichotomized into very good/good (good) and others (neither good nor poor/poor/very poor).
Int. J. Environ. Res. Public Health 2012, 9
3468
2.3. Independent Variables 2.3.1. Individual-Level Characteristics The participants indicated their education level by selecting one of several options, with regard to the highest grade completed. The education level was then categorized as: (1) None (no education), (2) Grades 1–7, (3) Grades 8–12, or (4) Higher than Grade 12. The respondents were also asked about their current employment status by asking them to pick one of several options. The options were then collapsed into three categories, namely: (1) Employed, (2) Unemployed, and (3) Permanently sick/student/pensioners/housewife not looking for job. Subjective socioeconomic status or social position was assessed on a continuous scale using responses to the following question asked in the 2007 SASAS (as in previous studies [14,15]): “In our society there are groups which tend to be towards the top and groups which tend to be towards the bottom. Where would you put yourself on a scale of 1 to 10, where 10 is the top and 1 the bottom?” Participants were asked to indicate if they had used dental services in the past year. Considering that utilization of dental services and self-rated oral health could be influenced by the presence of physical symptoms, participants’ recent history of oral health problems was assessed by asking them whether if, within the previous month, they had experienced any of the following common oral conditions: (1) bleeding gums when brushing (symptoms indicative of gum disease), (2) teeth sensitive to heat or cold, (3) bad breath, or (4) none. Participants were also asked how frequently they brush and to indicate if they use additional oral hygiene aids such as flossing and mouthwash. Current smokers and snuff users were those who indicated using the respective products “Every day” or “Some days”. Similar to the approach used in a previous study [16], trust was measured by asking respondents the extent to which they believed people could be trusted (trust is a proxy measure for social capital.). 2.3.2. Area-Level Characteristics Using the 2005 GHS data, area-level social network potential was derived from responses to the question “Is there a cellular telephone available to this household for regular use?” The response was either “Yes” or “No”. The aggregate percentage of cell-phone availability per household or the ‘cell phone network density’ of each municipal area was calculated and assigned to the respective municipal area where respondents to the 2007 SASAS resided. Also, using the 2005 GHS, the households that indicated that they did not consult with a health worker were asked why they did not consult any health worker during the past month, and five options were provided, including the option “not necessary”. For this analysis, the responses were dichotomized into two groups of respondents, namely, those who indicated they had experienced some or other barrier to accessing health services (1), and those who indicated experiencing no barrier (0). The proportion of those who had experienced a barrier in accessing health services were calculated for each municipal area as a proxy measure for the level of relative ease of access to health services among those living in that municipality. The participants were asked about the household’s main source of water and the respondents had to choose one of many options. Their responses were dichotomized into piped and non-piped water
Int. J. Environ. Res. Public Health 2012, 9
3469
sources, because piped water in South Africa has a relatively lower fluoride level (no public water fluoridation) than non-piped water sources, such as ground water [17]. The proportion of households without piped water was then computed for each area. A question about the main source of energy/fuel for the household was also asked. Like the other questions, it had many options which were collapsed into two: (1) those whose main energy source was electricity, and (2) those whose main energy source was not electricity. The proportion of households whose source of energy was not electricity was then computed for each area. For the purpose of analysis, each of these municipal or community-level variable values were auto ranked into three categories, namely those lower than the 33.3th percentile, those within the 33.3th to 66.7th percentiles and higher than the 66.7th percentile. 2.4. Data Analyses The data was analyzed using STATA version 10 (STATA Corp Inc., College Station, TX, USA). Group differences were tested using chi-squares and the t-test for categorical and continuous variables respectively. All statistical tests were two-tailed and the level of significance was set at p < 0.05. A multilevel binomial logit link model was used to assess the effect of community-level factors on self-rated oral health after the individual-level factors had been controlled for [18]. The main outcome variable was good self-rated oral health, and three sequential models were generated [19]. Model 1 was the empty model, which contained only the outcome variable with no independent variable. Area-level variables were added to Model 1 to become Model 2, and then the individual-level variables were added to Model 2 to become Model 3, which was the final model. The variance in self-rated oral health at the community level was noted for each of the models and the changes in these variance estimates were recorded as the model was built from an empty model to sequentially include the area-level factors and individual-level factors. These changes were noted to denote the level of contribution made by each set of factors/variables toward explaining variations in self-rated oral health across municipalities. The criterion for inclusion of the variables into the logistic model from the bivariate analysis was set at p < 0.25, while the decisive criterion for retention in the model was p < 0.05 [20]. Following a suggestion by Hosmer and Lemeshow [20], factors that did not meet the p < 0.25 criterion were finally introduced into the model to identify factors that were not significantly related to self-rated good oral health in themselves, but that made an important contribution in the presence of other variables. The log-likelihood ratio test (LR-test) was then used to examine whether or not the multilevel/random effect model was significantly better than an ordinary logistic regression model. Considering the previously noted modifying effect of gender on the observed association between level of education and self-rated oral health [9], the interactions between gender and education as well as between gender and social positioning as a measure of socioeconomic status were explored. As a result, additional analyses were carried out separately for male and female respondents. The interaction between gender and indicators of social capital was also explored based on the findings from other studies showing gender differences in the relationship between health and social capital [21,22].
Int. J. Environ. Res. Public Health 2012, 9
3470
3. Results and Discussion 3.1. Results The average age of the study participants was 36.9 (SD = 0.6) years. Of the respondents, 76.3% (n = 2,067) reported good oral health, 51.7% (n = 1,617) were female and 77.1% (n = 1,755) identified themselves as black Africans. The unemployment rate among the studied participants varied significantly by race, with black Africans having the highest unemployment rate. A significantly greater proportion of respondents in the age group between 16 and 24 years than respondents who were 65 years and older rated their oral health as good (Table 1). Those who reported dental attendance in the past year were less likely to have rated their oral health as good than those who did not. Other socio-demographic and behavioral factors associated with self-rated good oral health are shown in Table 1. Table 1. Bivariate relationship between self-rated good oral health status and individuallevel risk factors. Self-rated good oral health % (n)
Characteristics Socio-demographic factors Gender
0.00 Male Female
80.6 (939) 71.8 (1,189)
Ethnicity
0.03 Black Coloured Indian or Asian White
75.6 (1,315) 69.5 (278) 87.7 (273) 81.5 (262)
Age (Years)
0.00 16–24 25–34 35–44 45–54 55–64 >65
89.8 (604) 84.6 (523) 77.3 (474) 65.7 (270) 53.4 (154) 38.1 (96)
Education
0.00 >Grade 12 Grade 12 65
1.0 (Reference category) 0.67 (0.46–0.97) 0.48 (0.33–0.69) 0.24 (0.16–0.34) 0.16 (0.11–0.24) 0.12 (0.08–0.19)
Education >Grade 12 Grade 12 65 >Grade 12 Grade 12 65 >Grade 12 Grade 12